Discussion:
"IETF" working group for NNTP being created
(too old to reply)
Tulanet Management
2020-03-15 17:40:19 UTC
Permalink
Hello all,

A couple of people were interested in updating the NNTP protocol in
2020, so I posted a proposal on the DISPATCH working group for the IETF.

If you would like to make the world's largest social media network
(Usenet) better, then email me and sign up for the mailing list and send
your message of approval.

Thanks.
Grant Taylor
2020-03-15 18:06:29 UTC
Permalink
Post by Tulanet Management
Hello all,
Hi,
Post by Tulanet Management
A couple of people were interested in updating the NNTP protocol in
2020, so I posted a proposal on the DISPATCH working group for the IETF.
I saw the discussion on the DISPATCH mailing list.
Post by Tulanet Management
If you would like to make the world's largest social media network
(Usenet) better, then email me and sign up for the mailing list and send
your message of approval.
Is the mailing list available through the normal IETF channels? Or do
we currently need to email you before said channels are available?
--
Grant. . . .
unix || die
Tulanet Management
2020-03-15 21:40:25 UTC
Permalink
Post by Grant Taylor
Post by Tulanet Management
Hello all,
Hi,
Post by Tulanet Management
A couple of people were interested in updating the NNTP protocol in
2020, so I posted a proposal on the DISPATCH working group for the IETF.
I saw the discussion on the DISPATCH mailing list.
Post by Tulanet Management
If you would like to make the world's largest social media network
(Usenet) better, then email me and sign up for the mailing list and
send your message of approval.
Is the mailing list available through the normal IETF channels?  Or do
we currently need to email you before said channels are available?
The mailing list has not been "created" since the Working group hasn't
been officially created yet, but Russ says he has something already set
up so once we get off the ground we'll use that.

If you saw it, that's great, I was going to tell people who emailed me
to just subscribe to the list.
Russ Allbery
2020-03-15 21:46:24 UTC
Permalink
Post by Tulanet Management
The mailing list has not been "created" since the Working group hasn't
been officially created yet, but Russ says he has something already set
up so once we get off the ground we'll use that.
https://lists.eyrie.org/mailman/listinfo/ietf-nntp

was the mailing list for the former IETF working group and is still open.
Anyone who wants to talk about the future of the NNTP protocol, either for
a working group or just informally, is welcome to use it, and I'm happy to
offer it as the working group mailing list if/when a working group is
formed.
--
Russ Allbery (***@eyrie.org) <https://www.eyrie.org/~eagle/>

Please post questions rather than mailing me directly.
<https://www.eyrie.org/~eagle/faqs/questions.html> explains why.
Grant Taylor
2020-03-15 23:09:43 UTC
Permalink
Post by Tulanet Management
The mailing list has not been "created" since the Working group hasn't
been officially created yet, but Russ says he has something already set
up so once we get off the ground we'll use that.
ACK

Please add my name (and email address) as a someone interested in
participating.

I don't know how much help I'm qualified to provide, but I'd like to
contribute if at all possible.
Post by Tulanet Management
If you saw it, that's great, I was going to tell people who emailed me
to just subscribe to the list.
To clarify, I saw your emails to one of the IETF mailing lists, I
believe it was DISPATCH.

I remember seeing something else from Russ, but I've not looked at that
yet. $ToDoList++
--
Grant. . . .
unix || die
Tulanet Management
2020-03-15 23:17:20 UTC
Permalink
Post by Grant Taylor
Post by Tulanet Management
The mailing list has not been "created" since the Working group hasn't
been officially created yet, but Russ says he has something already
set up so once we get off the ground we'll use that.
ACK
Please add my name (and email address) as a someone interested in
participating.
I will keep you in mind if we create a new mailing list, but right now
it seems like we are using Russ' list for the time being, since the
working group hasn't been officially created.
Post by Grant Taylor
I don't know how much help I'm qualified to provide, but I'd like to
contribute if at all possible.
You are more qualified then you think :) Your ideas are what matter
Post by Grant Taylor
Post by Tulanet Management
If you saw it, that's great, I was going to tell people who emailed me
to just subscribe to the list.
To clarify, I saw your emails to one of the IETF mailing lists, I
believe it was DISPATCH.
Yes it was.
Post by Grant Taylor
I remember seeing something else from Russ, but I've not looked at that
yet.  $ToDoList++
$ToDoList-- (for me LOL)
n***@zzo38computer.org.invalid
2020-03-15 22:17:57 UTC
Permalink
Post by Tulanet Management
Hello all,
A couple of people were interested in updating the NNTP protocol in
2020, so I posted a proposal on the DISPATCH working group for the IETF.
If you would like to make the world's largest social media network
(Usenet) better, then email me and sign up for the mailing list and send
your message of approval.
I don't know what is that mailing list, and I certainly would think that
NNTP is much better than mailing list (and web forums!).

I think that NNTP is mostly OK how it is, although I do have some ideas:

* Increase article numbers to 63-bits. My own software already supports
this. That should hopefully be enough bits, and will work with existing
software before article numbers reach the old maximum of 31-bits.

* WHOAMI command to tell you what it would add to the Injection-Info
header. This would be an optional feature.

* ASCII should be preferred if possible (UTF-8 can be used if ASCII is
not suitable for some reason). I think that whether a question mark in a
wildmat matches a single byte or a single character should be
implementation-dependent. The same would be true of an asterisk. Note that
if no question marks are used and the text is valid UTF-8, then it will
still have the same meaning as before; also, if no non-ASCII is used in
the server, then the result is also the same. (Unicode is also not a
suitable character set for all purposes anyways. In order of most to least
preference, they should probably be: ASCII (best), UTF-8, others which are
compatible with principle of extended ASCII (worst); character sets which
are not compatible with principle of extended ASCII should be prohibited.)

* Convention for cross-posting between servers; it is not know how (or if)
this should work, although sometimes some users might want it if they are
using multiple servers and none of them have all of the wanted newsgroups.
My own software currently allows it as long as the client has specified
the message ID and at least one wanted newsgroup is known to the server.
However, this probably isn't the best way.
--
Note: I am not always able to read/post messages during Monday-Friday.
Tulanet Management
2020-03-15 23:24:32 UTC
Permalink
Post by n***@zzo38computer.org.invalid
Post by Tulanet Management
Hello all,
A couple of people were interested in updating the NNTP protocol in
2020, so I posted a proposal on the DISPATCH working group for the IETF.
If you would like to make the world's largest social media network
(Usenet) better, then email me and sign up for the mailing list and send
your message of approval.
I don't know what is that mailing list, and I certainly would think that
NNTP is much better than mailing list (and web forums!).
* Increase article numbers to 63-bits. My own software already supports
this. That should hopefully be enough bits, and will work with existing
software before article numbers reach the old maximum of 31-bits.
Yes, I think you meant 64-32. Good idea and I brought that point up.
Post by n***@zzo38computer.org.invalid
* WHOAMI command to tell you what it would add to the Injection-Info
header. This would be an optional feature.
Also a good idea as an optional feature.
Post by n***@zzo38computer.org.invalid
* ASCII should be preferred if possible (UTF-8 can be used if ASCII is
not suitable for some reason). I think that whether a question mark in a
wildmat matches a single byte or a single character should be
implementation-dependent. The same would be true of an asterisk. Note that
if no question marks are used and the text is valid UTF-8, then it will
still have the same meaning as before; also, if no non-ASCII is used in
the server, then the result is also the same. (Unicode is also not a
suitable character set for all purposes anyways. In order of most to least
preference, they should probably be: ASCII (best), UTF-8, others which are
compatible with principle of extended ASCII (worst); character sets which
are not compatible with principle of extended ASCII should be prohibited.)
Why ASCII? UTF-8 is backward compatible and it supports all languages +
emojis.
I don't think the NNTP protocol ever dictated what should be put in an
article (hence the yEnc/uuEncode/MIME wars a long time ago).
Post by n***@zzo38computer.org.invalid
* Convention for cross-posting between servers; it is not know how (or if)
this should work, although sometimes some users might want it if they are
using multiple servers and none of them have all of the wanted newsgroups.
My own software currently allows it as long as the client has specified
the message ID and at least one wanted newsgroup is known to the server.
However, this probably isn't the best way.
I don't really get what you are trying to say, but if I understand you
correctly, some Usenet servers auto create a NG if you either select the
article by Message ID and then select the NG that previously did not
exist. Works on the Abavia (XSUsenet) backbone, and Usenet.Farm
backbone. Suprisingly it does not auto create on Highwinds/Eweka/Omicron
servers, which are to date the largest Usenet operation in the world. At
least for the group alt.coronavirus, which I tried to create.

(without a newgroup message, I admit - I should probably send one)
:)
Adam H. Kerman
2020-03-16 03:34:30 UTC
Permalink
Post by Tulanet Management
Post by n***@zzo38computer.org.invalid
. . .
Why ASCII? UTF-8 is backward compatible and it supports all languages +
emojis.
An emoji certainly doesn't belong in a plain text article.

UTF character sets include ASCII. Obviously they are NOT backwards
compatible.
Post by Tulanet Management
I don't think the NNTP protocol ever dictated what should be put in an
article (hence the yEnc/uuEncode/MIME wars a long time ago).
As long as I write in English, I refuse to add MIME headers. I do so
only when I need to use a non-ASCII character.

You're right. It's not an NNTP issue. It's a USEFOR issue which I am
violating as we speak.

The problem really has nothing to do with Usenet. It's what the terminal
emulation is going to do. Do any terminal emulations change the translation
based on the received MIME headers? PuTTY doesn't. When I use an X-TERM,
it doesn't.

And what if the newsreader can't be bothered to parse the character set
in use and set the MIME headers based on the user's setting and not the
contents of the article? Happens all the time with followups.

Finally, when non-ASCII characters are used in headers, that is NOT
based on MIME headers. There are newsreaders out there that DO NOT parse
for non-ASCII characters in headers.

Here's a caveat to bear in mind. It doesn't matter what the standard
says. There will never come a time that a conventional article in plain
text can ever be posted to Usenet via Google Groups. There are certain
newsreaders that have never been capable of composing a conventional
article, ignoring the conventions of the eras they were written for.
Many newsreaders were abandoned decades ago but people continue to use
them. A new standard doesn't force anyone to write new code.
Post by Tulanet Management
Post by n***@zzo38computer.org.invalid
* Convention for cross-posting between servers; it is not know how (or if)
this should work, although sometimes some users might want it if they are
using multiple servers and none of them have all of the wanted newsgroups.
My own software currently allows it as long as the client has specified
the message ID and at least one wanted newsgroup is known to the server.
However, this probably isn't the best way.
I don't really get what you are trying to say,
He's talking about regional or institutional or local newsgroups, versus
international newsgroups. (By definition, a local newsgroup isn't
"Usenet" anyway.) These are all actual newsgroups with syntactically
correct names created on somebody's server, but not everybody's server
has the same set of newsgroups. He's saying he wants a user to be able
to crosspost even if not all the groups in the crosspost were created
locally, but he wants the group names parsed for correct syntax, which I
think is noble but hopeless.
Post by Tulanet Management
but if I understand you
correctly, some Usenet servers auto create a NG if you either select the
article by Message ID and then select the NG that previously did not
exist. Works on the Abavia (XSUsenet) backbone, and Usenet.Farm
backbone. Suprisingly it does not auto create on Highwinds/Eweka/Omicron
servers, which are to date the largest Usenet operation in the world. At
least for the group alt.coronavirus, which I tried to create.
(without a newgroup message, I admit - I should probably send one)
:)
As we discussed before, please DON'T name it into the second level. I'd
suggested better names for you to use.
Michael Bäuerle
2020-03-16 10:38:49 UTC
Permalink
Post by Adam H. Kerman
[...]
As long as I write in English, I refuse to add MIME headers. I do so
only when I need to use a non-ASCII character.
You're right. It's not an NNTP issue. It's a USEFOR issue which I am
violating as we speak.
The problem really has nothing to do with Usenet. It's what the terminal
emulation is going to do. Do any terminal emulations change the translation
based on the received MIME headers? PuTTY doesn't. When I use an X-TERM,
it doesn't.
This is what I would expect. If I start an xterm, it uses the encoding
of my locale (and doesn't know what MIME is).

It's the job of the newsreader to translate received MIME encoded
content to whatever the local system is configured for (in this case
to what LC_CTYPE says).


[Fup2 set to news.software.readers, because it's not transport related]
Adam H. Kerman
2020-03-16 14:37:44 UTC
Permalink
Post by Michael Bäuerle
Post by Adam H. Kerman
[...]
As long as I write in English, I refuse to add MIME headers. I do so
only when I need to use a non-ASCII character.
You're right. It's not an NNTP issue. It's a USEFOR issue which I am
violating as we speak.
The problem really has nothing to do with Usenet. It's what the terminal
emulation is going to do. Do any terminal emulations change the translation
based on the received MIME headers? PuTTY doesn't. When I use an X-TERM,
it doesn't.
This is what I would expect. If I start an xterm, it uses the encoding
of my locale (and doesn't know what MIME is).
It's the job of the newsreader to translate received MIME encoded
content to whatever the local system is configured for (in this case
to what LC_CTYPE says).
[Fup2 set to news.software.readers, because it's not transport related]
Please don't screw with followups, as this whole discussion has been
crossposted from the start. I personally object to most crossposted
discussion but I'm not the one who started this thread. Ideally you
would have started a SEPARATE thread that's not crossposted in lieu of
following up to my article. In this case, we should maintain the
crosspost throughout this thread.

My newsreader passes along characters received as is. If it receives QP
encoding, it calls an outside process to decode, but then the resulting
character to display won't display as intended if there's a mismatch
with locale if it's 8 bit, given that LC_CTYPE is set for UTF-8. Decoding
is performed in the terminal. I have different xterms preset to display
character sets as received. If I have to use non-ASCII characters, I just
add MIME headers to the article, but if I'm able to, I just substitute
ASCII for non-ASCII in the quoted text.
Julien ÉLIE
2020-03-16 18:08:51 UTC
Permalink
Hi Adam,
Post by Adam H. Kerman
I personally object to most crossposted
discussion but I'm not the one who started this thread. Ideally you
would have started a SEPARATE thread that's not crossposted in lieu of
following up to my article. In this case, we should maintain the
crosspost throughout this thread.
A point to discuss for a possible integration into USEAGE:
https://tools.ietf.org/html/draft-ietf-usefor-useage-01
--
Julien ÉLIE

« Passion is inversely proportional to the amount of real information
available. » (Benford's law)
Adam H. Kerman
2020-03-16 18:37:11 UTC
Permalink
Post by Julien ÉLIE
Hi Adam,
Post by Adam H. Kerman
I personally object to most crossposted
discussion but I'm not the one who started this thread. Ideally you
would have started a SEPARATE thread that's not crossposted in lieu of
following up to my article. In this case, we should maintain the
crosspost throughout this thread.
https://tools.ietf.org/html/draft-ietf-usefor-useage-01
Thanks for bringing that up. I've been meaning to ask if it would be
worthwhile to complete work on USEAGE. Did those working on that
document have a mailing list separate from the list that Russ offered?
Russ Allbery
2020-03-16 19:12:23 UTC
Permalink
Post by Adam H. Kerman
Thanks for bringing that up. I've been meaning to ask if it would be
worthwhile to complete work on USEAGE. Did those working on that
document have a mailing list separate from the list that Russ offered?
Yeah, it was discussed on the USEFOR list, which I didn't host. I'm not
sure if that one is still open.
--
Russ Allbery (***@eyrie.org) <https://www.eyrie.org/~eagle/>

Please post questions rather than mailing me directly.
<https://www.eyrie.org/~eagle/faqs/questions.html> explains why.
Julien ÉLIE
2020-03-16 19:53:06 UTC
Permalink
Hi Russ,
Post by Russ Allbery
Post by Adam H. Kerman
Thanks for bringing that up. I've been meaning to ask if it would be
worthwhile to complete work on USEAGE. Did those working on that
document have a mailing list separate from the list that Russ offered?
Yeah, it was discussed on the USEFOR list, which I didn't host. I'm not
sure if that one is still open.
It is still open.
I wrote there last year (2019):
https://www.ietf.org/mailman/listinfo/usefor
--
--
Julien ÉLIE

« Passion is inversely proportional to the amount of real information
available. » (Benford's law)
Adam H. Kerman
2020-03-16 03:14:42 UTC
Permalink
Post by n***@zzo38computer.org.invalid
* WHOAMI command to tell you what it would add to the Injection-Info
header. This would be an optional feature.
Eh. Just send an article.
Post by n***@zzo38computer.org.invalid
* ASCII should be preferred if possible (UTF-8 can be used if ASCII is
not suitable for some reason). . . .
I try to explain to people that ASCII had limited internationalization.
Various characters were ambiguous and could be used for diacritics. In a
teletypewriter, the desired character was created with a strike,
backspace, overstrike sequence. Alas, terminals were inconsistent about
how these sequences were displayed and the ability to assemble a
character with the base character and diacritic was lost to
the ages. Also, backspace is typically used for RUBOUT anyway thanks to
DOS. Nearly all of us map it to ^H.
Post by n***@zzo38computer.org.invalid
* Convention for cross-posting between servers; it is not know how (or if)
this should work, although sometimes some users might want it if they are
using multiple servers and none of them have all of the wanted newsgroups.
My own software currently allows it as long as the client has specified
the message ID and at least one wanted newsgroup is known to the server.
However, this probably isn't the best way.
I don't think that's a network issue, whether the injecting server allows
a crosspost to a group not created locally. Furthermore, that creates all
sorts of problems if the user misspelled the group name or just got the
group name wrong, like thinking that the group was named as the third
level rather than the fourth level, or thinking there's a Big 8 group
when it's actually an alt.* group. Chris Caputo will create a group from
a crossposted article even if there is a syntactically correct name for
the group the author probably wanted. There's just no way to tell what's
the syntactically correct name except for what's in the active file.
Note that the active file at ftp.isc.org isn't comprehensive anyway, as
that's updated only if the newgroup message included a For your
newsgroups file line.

Most crossposts aren't helpful.
n***@zzo38computer.org.invalid
2020-03-16 04:49:48 UTC
Permalink
Post by Adam H. Kerman
I don't think that's a network issue, whether the injecting server allows
a crosspost to a group not created locally. Furthermore, that creates all
sorts of problems if the user misspelled the group name or just got the
group name wrong, like thinking that the group was named as the third
level rather than the fourth level, or thinking there's a Big 8 group
when it's actually an alt.* group. Chris Caputo will create a group from
a crossposted article even if there is a syntactically correct name for
the group the author probably wanted. There's just no way to tell what's
the syntactically correct name except for what's in the active file.
Note that the active file at ftp.isc.org isn't comprehensive anyway, as
that's updated only if the newgroup message included a For your
newsgroups file line.
OK, those are valid points; you are correct. All of the things you mention
are in fact possible problems, which I have not considered, but you are
correct that they will be some of the problems.
--
Note: I am not always able to read/post messages during Monday-Friday.
Grant Taylor
2020-03-17 04:30:04 UTC
Permalink
Post by Adam H. Kerman
Most crossposts aren't helpful.
I believe that judicious use of crossposts, preferably with proper
Followup-To: can be quite useful. Though sadly, it's too easy to
{,mis,ab}use crossposts. Much like Reply-All.

I've seen a good number of cross posted messages asking for help, not
knowing which group to post to (or the proper ideal group is largely
dead) and asking for help with follow ups to one group. As long as the
number of groups isn't too large, I find this to be acceptable.
--
Grant. . . .
unix || die
Adam H. Kerman
2020-03-17 05:51:26 UTC
Permalink
Post by Grant Taylor
Post by Adam H. Kerman
Most crossposts aren't helpful.
I believe that judicious use of crossposts, preferably with proper
Followup-To: can be quite useful.
Followup-To means "I know I'm posting off topic, but I'm telling the
next guy how to behave."

If the author has determined that his own followup is off topic in several
of the groups in the crosspost, then he's obliged to cut the crosspost.
Post by Grant Taylor
Though sadly, it's too easy to
{,mis,ab}use crossposts. Much like Reply-All.
I've seen a good number of cross posted messages asking for help, not
knowing which group to post to (or the proper ideal group is largely
dead) and asking for help with follow ups to one group. As long as the
number of groups isn't too large, I find this to be acceptable.
Michael Bäuerle
2020-03-19 13:02:10 UTC
Permalink
Post by Adam H. Kerman
Followup-To means "I know I'm posting off topic,
No. Annother case are the release announcements for my newsreader with
Xpost to this group and <news:de.comm.software.newsreader>. The reason
for the Followup-To header field there is not off topic, but simply that
most people can understand and write english (therefore it makes sense
that people, who speak both languages, use english and this group).
Post by Adam H. Kerman
but I'm telling the next guy how to behave."
"telling ... how to behave" sounds imperative. Because it cannot be
enforced anyway, it should be interpreted as a suggestion like:
"I think this should be discussed here, follow if you agree".
Post by Adam H. Kerman
If the author has determined that his own followup is off topic in several
of the groups in the crosspost, then he's obliged to cut the crosspost.
Done for this article (with new subject to indicate the new subthread).
But I'm not convinced that this is the better method in general.

Anonymous
2020-03-16 16:38:56 UTC
Permalink
Post by Tulanet Management
Hello all,
A couple of people were interested in updating the NNTP protocol in 2020, so
I posted a proposal on the DISPATCH working group for the IETF.
If you would like to make the world's largest social media network (Usenet)
better, then email me and sign up for the mailing list and send your message
of approval.
Thanks.
Sign up is a first border. What next?
Keep NNTP as is.
Loading...